more lunacy from the BBC

The E/M simulation of the un-mated half of a connector isn't slap-dash -

To your complete satisfaction.

We didn't try. We put SMA sockets on the printed circuit board and routed coax cable through a mixed signal 41812 connector. You can put through mixed signal D-type connectors too. Just to add high end appeal we used semi-rigid coaxial cable to link everything together

So why use FR4? We put in two layers of isocyanate bonded Teflon clothe as the top and bottom layers of our six-kayer board and that was back in

Few people deserve contempt. Some of my colleagues screwed up from time to time, and so did I, but we covered up for one another.

Wouldn't it have been a better-organised firm if you could have gone to the chief engineer and told him what you had done wrong, so that the knowledge could be passed on?

Instructions are passed down the management chain but management needs to have information passed back up to them, so they can make informed decisions.

We didn't need to bother. The technicians always knew when you had screwed up, and the word got passed around.

If that were the only way information got passed back up to them the firm would have foundered very rapidly. In general management doesn't want to know and wouldn't have known what it meant if they had been told through official channels. Unofficial channels tended to be more informative.

I did make a habit of talking to the service engineers from time to time, when they were back on-site. I didn't have to make an effort with the technicians - they loved to gossip.

Yes, I've never worked anywhere where there were formal channels, the informal one was the coffee break and the canteen and it worked very well. The Chief Engineer used to wander around the development department having regular chats with the design engineers to make sure everything was running smoothly and to step in if necessary.

That's a good plan - and even going out as a locum service engineer from time to time, so you really get the feel for what they have to contend with. In one firm I worked for, the designer of a piece of equipment was on call to the production line and the repair shop if any of his designs caused them problems.

That might take two months and cost a few hundred times as much.

This isn't bad at all, for a $2 JLC board.

One problem with teflon lams is that the copper adhesion is terrible. Vias can be flakey too. Some of the rigid Isola materials are better, if you really need a low-loss dielectric. For short traces sending logic levels around, an exotic lam isn't worth it. Even crazy fast PCIe is done on FR4.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

We don't have engineering technicians. We can solder ourselves.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

We had an actual production line. We did solder ourselves, and some of my get-it-to-work-as-it-was-designed-to-work changes were allowed to go out to customers.

It didn't, even in the UK in 1987. The boards were hideously expensive - the printed circuit house had had to buy quite a large minimum chunk of the isocyanate bonded Teflon cloth substrate to make our boards, and they got the money back over of a handful of boards

Ours were fine. Nowadays there a whole range of high frequency substrates, and we used couple of different ones even back in the late

It's not just the low-loss. The glass-fibre mat in FR4 means that the dielectric constant of the substrate changes along the transmission line, and that generate small reflections.

Keep the specification sloppy enough and you can get away with murder.

If you need good performance, you have to take more care.

Cambridge Instruments worked that way too. Between projects the design engineers put in time on "mods" - modifications proposed by the production technicians to make boards more reliable or easier to set up. About half of them were misconceived, but the other half could get quite interesting, and we did have to talk to the production line technicians which was good for informal communications.

You think the PCIe spec is sloppy?

Use something else in your PC.

Or adaptive equalization.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

John Larkin wrote: |-----------------------------------------------------------------| |"We don't have engineering technicians. We can solder ourselves."| |-----------------------------------------------------------------|

All engineers are technicians, as all technicians are engineers, by definitions. (S.

Liz Tuddenham snipped-for-privacy@PoppyRecords.invalid.invalid> wrote: |-----------------------------------------------------------------------| |"Bill Sloman snipped-for-privacy@ieee.org wrote: | | | | | |> ...Some of my colleagues screwed up from time | |> to time, and so did I, but we covered up for one another. | | | |Wouldn't it have been a better-organised firm if you could have gone to| |the chief engineer and told him what you had done wrong, so that the | |knowledge could be passed on? | | | |Instructions are passed down the management chain but management needs | |to have information passed back up to them, so they can make informed | |decisions." | |-----------------------------------------------------------------------|

Correct communications are good! (S.

No. Technicians seldom design things. By definition!

Engineers seldom do routine production test or repairs.

Engineers get paid more.

What do you do?

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Methinks you'll find that the salaries of electronic engineers and technicians are quite different. From a random AI dust bin of wisdom:

"In California, the average salary for an electronics technician is approximately $60,000-$66,000 annually ($29-$32 per hour) as of April

"The average salary for an electronics engineer in the U.S. is approximately $111,910 to $127,590 per year, according to 2024-2026 data. Top earners can exceed $140,000 annually..."

In other words: two technicians = one engineer

People hired as techs very rarely advance to engineering positions.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Well, apart from Jim Williams of LTC and Errol Dietz, who started out as Bob Pease’s tech and wound up as CTO of National Semi.

Cheers

Phil Hobbs

It's a mass market product, and the production volumes are going to be higher that 100,000 units per year. You don't seem to design for that kind of market, and I certainly never did. You can put in a lot of design effort to let you get away with cheapest possible solutions.

Why would I? It's a cheap mass market product, and it works. If I needed something better - and could afford it - I'd buy that.

Adaptive equalisation won't get rid of the small reflections from the lumps of glass fibre in FR4.

I met Jim once at the Foothill Flea Market. He was very reserved, very shy, but shared my affection for the old Tek 500-series scopes.

His two books on analog design are wonderful.

You can see from his schematics the lack of cademic engineering, especially control theory. He was very intuitive.

Jim and Pease died about the same time.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Of course it will.

Given a linearly distorted signal without too much noise, there is a reciprocal function that undoes the distortion. It's "the deconvolution problem" to find that reciprocal function.

The deconvolution thing is one of the mathematically "ill-posed problems." Which makes it fun to play with.

I can't see glass weaves in a 20 GHz TDR on a cheap FR4 board.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics